A leading automotive glass parts manufacturer supplying major OEMs faced fragmented demand inputs and manual planning bottlenecks. Learn how Arketic AI's demand interpretation engine automated requirement understanding, anomaly detection, and production estimation — reducing planning time to one-eighth of previous levels.
A leading automotive glass parts manufacturer supplies windshields, side windows, rear glass, and specialty glass components to major automotive OEM manufacturers. Operating in a just-in-time, high-precision supply chain, the company must align its production planning tightly with OEM requirements while managing material costs, capacity constraints, and delivery SLAs.
The company faced increasing complexity in coordinating production with multiple automotive OEM partners.
Fragmented Requirement Inputs: OEM demand forecasts and requirement lists arrived in different formats, frequencies, and levels of detail across various stakeholders.
Manual Interpretation and Planning: Planners manually translated OEM requirements into internal part breakdowns and ERP plans, a process that was time-consuming and error-prone.
Limited Forward Visibility: Weekly and multi-week demand fluctuations were difficult to anticipate accurately, leading to reactive rather than proactive planning.
Late Detection of Anomalies: Unexpected order spikes, drops, or mismatches between OEM demand and ERP data were often detected after they had already impacted production or inventory levels.
High Dependency on Key Planners: Critical know-how was locked in human experience, creating significant operational risk and single points of failure.
The company implemented Arketic AI's demand interpretation and production estimation engine, integrated with OEM interfaces and the internal ERP system. The solution delivered the following core capabilities:
Automated Requirement Understanding: AI ingests OEM requirement lists from multiple automotive manufacturers, finding relevant data through document processing or API connections. The system normalizes formats, terminology, and units across OEMs, and maps requirements directly to the company's internal part and BOM structure.
Intelligent Production Estimation: The engine combines OEM demand with ERP master data, inventory levels, production capacity, and historical demand patterns to produce rolling weekly part breakdown forecasts.
Human-in-the-Loop Approval: Arketic AI generates recommendations, not blind automation. Planners review, adjust, and approve forecasts before ERP posting, ensuring trust, accountability, and regulatory compliance. Planners can intervene at any point, analyze data, and set milestones in the automation and analysis of forecasts.
Anomaly Detection and Classification: Arketic AI continuously monitors the end-to-end flow and flags anomalies directly into the ERP, including unexpected quantity spikes or drops; OEM demand deviating from historical patterns; mismatches between OEM demand and ERP stock levels; unusual part mix changes such as sudden model-specific shifts; capacity or material constraint conflicts; and timing inconsistencies in demand signals. Each anomaly is classified, explained, and prioritized, enabling faster human response.
Automotive OEM manufacturers have different means of sharing their BOM lists: some via documents posted on FTP servers, some through APIs, and even via email. The key integration challenge was funneling this data into an organized list of BOMs for the company's own operational processes.
While a simple integration might not appear to require AI, the real value emerged in detecting incorrect file formats or contents, errors in input data not visible to the human eye, and unexpected data patterns based on seasonality. Arketic AI performs these checks based on previous orders and operational history, providing alerts to the relevant teams.
The system acts as an operational logic engine that incorporates past seasonal data, a task that is not only difficult but also involves sensitive or even regulated information that cannot easily be disclosed to cloud-based AI providers for such analysis.
Data is fed into the company's ERP and MES systems for production execution. Thanks to Arketic AI's workflow-based capabilities, an optimized operations analysis and flow is seamlessly realized, with continuous reporting available throughout.
Arketic AI's operations engine parses and optimizes processes to one-eighth of the time previously required using standard approaches such as RPA. While RPAs typically operate based on fixed rule sets and KPIs requiring pre-set anomaly and reaction models, Arketic AI makes decisions on the fly, always with a final human approval step.
The project utilized Arketic AI's own LLM models fed through its RAG-enabled architecture, processing sensitive production historical data without touching the internet for security and compliance. In-company knowledge assets were leveraged without sharing sensitive trade data to cloud-based AI engines, delivering a best-of-breed solution.
Previously, erroneous data fed by OEM partners not only consumed valuable planning staff time but also impacted production planning and delivery SLAs, the cost of which is difficult to quantify. Arketic AI compressed this error resolution time to minutes, saving time while helping meet contractual obligations.
Following the highly successful implementation, future plans include: raw material and supplier integration to extend forecasting to glass raw materials and upstream suppliers; energy and cost optimization to align production forecasts with energy price predictions; quality and scrap correlation to link production anomalies with defect rates; cross-plant optimization to balance demand across multiple facilities using AI-driven load distribution; and a digital twin of production planning to simulate what-if scenarios for OEM changes, capacity limits, or disruptions.
Additional plans include standard HR, financial, and industry-imposed standards adherence and reporting cycle implementations across the organization.
Conventional solutions typically fall into one of the following categories: acting as a hub for one or multiple AI solutions; using external robotic engines to automate flows once AI output is generated; or requiring expert AI engineers to tune and integrate into existing environments.
Arketic AI stands apart by enabling organizations to incorporate their sensitive existing data without disclosing it to third-party cloud providers. AI decisions rely not only on publicly available web data but also on the company's own proprietary data, all without sending sensitive information outbound.
Architecturally, Arketic AI can function on-premises or in Arketic's own cloud data center, fully compliant with local and national regulations for holding sensitive customer data.
With its integrated automation engine, Arketic AI enables seamless flow of AI-generated outputs without the need for any external third-party toolset. And through its agent creation engine with straightforward tutorials, users can create their own agents without requiring Arketic professional services.
%20(184%20x%2044%20piksel).svg)
